Wan Weiwei, a researcher from Osaka University, Japan, introduced the rapid integration method and application of robot based on WRS system

Intelligent robot has increasingly become an indispensable subject in all kinds of production , Through system integration , It can form various application systems with its flexible and efficient characteristics , Increased productivity , Enhanced flexibility , Adapted to today's development trend .

《 Automation and system integration in the era of intelligent robots 》

Robot operating system is an operating system in which robots are the main body of processing and service , The robot is just a part of the whole operating system , All systems should be integrated to form an organic whole , To complete the task , Meet the needs of the job . The limitations of traditional system integration methods fail to give full play to the advantages of intelligent robots , This leads to increased specificity of the robot system , Therefore, it breaks through the traditional system integration methods , It is of great significance to establish a new robot operating system .

This paper discusses the automation and system integration technology of intelligent robot ,NOKOV Measurement motion capture invites its Japanese customers – Associate Professor, School of basic engineering, Osaka University, Japan , Professor wanweiwei, researcher of symbiotic intelligent system research center of Osaka University , United robot lecture hall , Share relevant knowledge .

The key points shared in this issue are as follows

Introduction to intelligent robot planning and control system

be based on WRS Rapid integration method of robot system

be based on WRS Rapid robot integration application of the system

The following content is sorted out according to Professor Wan Weiwei's live sharing record

01 Introduction to intelligent robot planning and control system

Traditional robot system integration methods mainly include teaching board 、 Traction and virtual space simulation environment , These methods have the following problems ： First, it can not be applied to different manipulator hardware equipment ; Second, it cannot be applied to the production environment of variant variables .

In recent years , Disadvantages of traditional integration methods , Some solutions have been proposed , Then there are some new start-ups , Its technical basis lies in visual recognition and motion planning , It is mainly used for handling 、 assemble 、 Carry 、 machining 、 Assembly, etc , It requires the cooperation of multiple sensors including force sensors to achieve .

Based on the existing technology and development experience , The laboratory considered the following framework for robot system integration ：

• Teaching completely separated from robot
• Visual recognition and simultaneous planning of tasks and actions , Use reinforcement learning to solve the search and optimization problems that are difficult to solve by traditional methods
• Multimode sensor feedback , Error identification and recovery , Adjust the error gently through various sensors , Improve the robustness of the system

Based on the above frameworks , The laboratory is from 2016 An open source software system has been developed since ——WRS, The software is based on ROS、RTM、Choreonoid、OpenRAVE Development , And ROS similar , Also considered URDF The definition structure of , Including independent R & D planning 、 control 、 Communication and vision modules , It is characterized by full open source and continuous updating iteration 、 In innovation .

02 be based on WRS Rapid integration method of robot system

How to understand the development process of robot system ？ Through one WRS The case of system development .

Take the robot that replaces the workers working in front of the workbench as an example , The development process is roughly divided into the following three steps ：1、 Visual recognition , Used to identify the position and posture of objects , At present, there are many library functions ;2、 Grasp the planning , The curvature of the surface of an object 、 Roughness 、 Bending and possible collision of manipulator , Make the manipulator grasp with a reasonable posture ;3、 Integrated Mission and action planning , Its core is to optimize its movement , To achieve the effect that no operation is required after only issuing an instruction .

Through the above methods, people can “ teaching ” As a result, the end-to-end operation of the robot , Solve the problem that robots can not be used universally 、 Variable production, etc . In this development mode , There is no need for people to design robot actions , The robot can “ decision ” How to do the action , It can even perform assembly operations such as hole alignment without sensor control , For example, using geometric constraints to avoid force control , So as to avoid damage to the mechanism and sensor .

03 be based on WRS Rapid robot integration application of the system

We can learn from the following examples based on WRS Application of rapid integration method of robot under the system .

1、 From the position “ teaching ” To robot action . Several objects can be stacked manually by people , Set goals for the robot , Then the robot can according to the target , Automatically design actions and timely feedback and correction , Finally, the same stacking effect is achieved .

2、 feedback 、 Error detection and error recovery . For example, assign the screw position to the robot to make it work automatically , Robots can independently design actions for subsequent industrial production .

3、 Force control oriented planning optimization . Effective insertion and assembly action can be carried out through force feedback .

4、 Use geometric constraints to avoid force control . The use of force control may cause certain workpiece damage , For precision equipment assembly with high surface requirements , It may be more appropriate to use visual feedback at this time . The geometric constraint method can make the robot obtain the exact geometric position of the object conveniently , A relatively simple visual inspection method can be used to efficiently realize positioning assembly without force feedback .

5、 Mixed planning of tasks and actions . Take an operation of stacking wood as an example , Robots can be in the process of stacking , Carry out hand holding operation , To keep the assembly from collapsing .

6、 Designing tools for robots . In order to improve the universality of robots , For the end effector of the robot , Replacement should be minimized , such , In order to realize the fine operation under different work tasks , Teacher Wan and his lab have designed various special tools to match the fixed manipulator , For example, in the figure below , The two hands of the robot hold suction cups and clamps respectively , It can perform a variety of operations .

in addition , A tool device is also designed to convert the lateral pressing motion of the robot manipulator into rotation , In this way, it is easy to screw in and other operations without adding other electrical equipment .

7、 Three dimensional mapping through feedback and error detection , In this way, you can draw on the surface of buildings or other objects whose surface is not completely flat .

8、 By identifying 、 planning 、 Force control and so on complete the work as complex as human beings . For example, use pulley block and rope to turn over the slate .

9、 Man machine cooperation . Robots do what robots are good at , What people are good at , This can greatly improve work efficiency .

Source of the article ： leaderobot